Recent studies have reported a link between gut microbiota and nonalcoholic fatty liver disease (NAFLD), showing that germfree (GF) mice do not develop metabolic syndromes, including NAFLD. However, the specific bacterial species causing NAFLD, as well as their molecular cross talk with the host for driving liver disease, remain elusive. Here, we found that nonvirulent endotoxin-producing strains of pathogenic species overgrowing in obese human gut can act as causative agents for induction of NAFLD and related metabolic disorders. The cross talk between endotoxin from these specific producers and the host’s TLR4 receptor is the most upstream and essential molecular event for inducing all phenotypes in NAFLD and related metabolic disorders. These nonvirulent endotoxin-producing strains of gut pathogenic species overgrowing in human gut may collectively become a predictive biomarker or serve as a novel therapeutic target for NAFLD and related metabolic disorders.
Gut microbiota-derived endotoxin has been linked to human nonalcoholic fatty liver disease (NAFLD), but the specific causative agents and their molecular mechanisms remain elusive. In this study, we investigated whether bacterial strains of endotoxin-producing pathogenic species overgrowing in obese human gut can work as causative agents for NAFLD. We further assessed the role of lipopolysaccharide (LPS)-Toll-like receptor 4 (TLR4) cross talk in this pathogenicity. Nonvirulent strains of Gram-negative pathobionts were isolated from obese human gut and monoassociated with C57BL/6J germfree (GF) mice fed a high-fat diet (HFD). Deletion of waaG in the bacterial endotoxin synthetic pathway and knockout of TLR4 in GF mice were used to further study the underlying mechanism for a causal relationship between these strains and the development of NAFLD. Three endotoxin-producing strains, Enterobacter cloacae B29, Escherichia coli PY102, and Klebsiella pneumoniae A7, overgrowing in the gut of morbidly obese volunteers with severe fatty liver, induced NAFLD when monoassociated with GF mice on HFD, while HFD alone did not induce the disease in GF mice. The commensal Bacteroides thetaiotaomicron (ATCC 29148), whose endotoxin activity was markedly lower than that of Enterobacteriaceae strains, did not induce NAFLD in GF mice. B29 lost its proinflammatory properties and NAFLD-inducing capacity upon deletion of the waaG gene. Moreover, E. cloacae B29 did not induce NAFLD in TLR4-deficient GF mice. These nonvirulent endotoxin-producing strains in pathobiont species overgrowing in human gut may work as causative agents, with LPS-TLR4 cross talk as the most upstream and essential molecular event for NAFLD.